Half the Jobs You Are Looking At Will Be Gone in 13 Days

A few weeks ago someone in our Discord mentioned they were saving roles into a spreadsheet and doing a proper application push every other Sunday. Tidy system. Better than most people manage.

It also meant that by the time they sat down to apply, half of what they had saved no longer existed.

We had never actually measured that, which is embarrassing for a company whose whole job is watching career pages. So we measured it.

The number

We looked at 11,219 roles that opened and then closed at companies we were already tracking, and timed how long each one stayed up.

Half of tech roles are gone thirteen days after they appear. Just under a third (29.6%) are gone inside a week. By day 30, 71.6% of them have closed.

The long tail is real but thin: 3.6% stay open past 90 days. Those are the ones you remember, because they are still sitting there the next time you look, which makes them feel representative. They are not.

Why the average is a lie

Notice the gap between the median (13) and the mean (24). That is a distribution with a fat right tail, and it is the reason "jobs are usually open for about a month" feels true and is useless.

A handful of roles stay open for a quarter or more. Perpetual reqs, hard-to-fill infrastructure roles, teams that would hire the right person any time but are not urgently short-staffed. They drag the average up by eleven days and they are the ones least likely to be your role.

The median is the number that should change your behaviour. Thirteen days.

What that does to a job search

Run the arithmetic on the every-other-Sunday plan. Fourteen days between sessions, median lifespan thirteen days. On average you are applying to roles that are already closed, and you will never see the ones that opened and closed in between.

That is not a discipline problem. That person was more organised than most. The cadence was just tuned to a market that closes faster than the cadence.

A few things follow, and none of them are "apply to more jobs":

Freshness beats volume. A role you see on day two has a materially different chance than the same role on day twenty. Same job, same resume, different queue position. This is the single cheapest edge available and it costs nothing but timing.

Weekly is the slowest cadence that works. Below that you are systematically sampling the leftovers. (Daily is better. Daily is also how people burn out by week three, so weekly is the honest recommendation.)

A saved list decays. Whatever you are keeping roles in, assume half of it is stale after two weeks. The spreadsheet is not wrong, it just needs a shorter shelf life than most people give it.

Where this gets uncomfortable for us

Here is a thing we did not enjoy finding.

We only mark a role closed after it has gone missing from a career page three scrapes in a row. That grace period exists because career pages flicker, and we would rather be slow than tell you a live job is dead.

But it means every number above is measured to our close date, not the employer's. The real lifespans are shorter than what we published. We do not know by exactly how much, and we are not going to guess at a correction, but the bias runs one direction only: thirteen days is the generous version.

There is a second thing worth naming. We time from when we first see a role, not from when it was posted. If a company puts a job up on Monday and our scraper reaches them Wednesday, we lose two days. To keep that from turning into "we measured our own onboarding speed," this analysis only counts roles that appeared at least fourteen days after we started tracking that company, so every role in it went up while we were already watching. That threw out roughly a third of the data and it was worth it.

Both corrections push the same way. Faster than we said.

What we do not know

Whether a closed role was filled. We watch career pages, not offer letters. A role that disappears on day six might have been filled from an internal referral before the post was ever real, or the headcount might have been pulled in a budget meeting. From outside, both look identical, and anyone claiming to tell them apart from scraped data is guessing.

That matters for how you read the fast tail. Some of that 29.6% closing inside a week was never a fair fight. We just cannot tell you which part.

We also cannot break this down by seniority yet with any confidence, which is genuinely the thing we most want to know. (Intuition says junior roles close fastest because they get buried in applications. Intuition has been wrong about our data before, so we are not printing a number until we trust it.)

The practical version

If you take one thing: the useful unit of a job search is the week, not the month. Whatever you have built, tighten it until it turns over weekly.

The reason we built Remoet around starring companies rather than saving jobs is more or less this post. A saved job has a thirteen day half-life. A company you care about keeps hiring for years, and when you star one you get its whole open list right away, not a wait for the next posting. You can star companies from the site, or connect your agent and have it check the ones you starred while you get on with your week.

Now the part where we ask for help. We want the seniority split, and we suspect several of you have a better instinct for what is going on in the fast tail than our data can show. If you have watched a role in your own company open and close in six days, we would love to know what actually happened to it. Come tell us on Discord.